What are Tri-21B-Think's benchmark scores?

As of 25 Oct 2026, Tri-21B-Think (thinking effort) scores 43.4 ±5.9 on the BenchLeader Index, strongest in instruction following (55) and weakest in knowledge (33). That rests on 6 independent benchmark results, and every score on this page links to its publisher.

How does Tri-21B-Think rank?

Tri-21B-Think (thinking effort) is #568 of 785 ranked configurations on 25 Oct 2026. 350 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.

When was Tri-21B-Think released?

Trillion Labs released Tri-21B-Think on 10 Feb 2026.

Trillion LabsOpen weightsReasoning modelCompare with another model

Tri-21B-Think

Tri-21B-Think is a Trillion Labs open-weights reasoning model, released 10 Feb 2026. Its best configuration (thinking reasoning effort) ranks #568 of 785 on the BenchLeader Index at 43.4 ±5.9, in the lower half. The ± is the point: 350 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in instruction following (55) and lowest in knowledge (33). Last measured 10 Feb 2026.stale: no new result in six months

Blended price: –
Output speed: –
First answer: –
Context: 32k
Full answer: –
Cost per run: –
Released: 10 Feb 2026

How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.

Overall index43
Reasoning41
Agents & tools36
Knowledge33
Instruction following55
Long context35
Composite40

How its index has moved

9 Sept 2026 to 21 Oct 2026

The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.

Benchmark results

See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Reasoning

Benchmark	thinking	Source
GPQA Diamond (AA)not in index	60.1%#376	Artificial Analysis
Humanity's Last Exam (AA)not in index	5.9%#411	Artificial Analysis
CritPt	0.3%#271	Artificial Analysis

Agents & tools

Benchmark	thinking	Source	Trend
Terminal-Bench Hard	2.3%#320	Artificial Analysis
τ²-Bench Telecom (AA)not in index	93.3%#43	Artificial Analysis

Knowledge

Benchmark	thinking	Source
AA-Omniscience	-62.3#478	Artificial Analysis
AA-Omniscience: accuracynot in index	12.3%#471	Artificial Analysis
AA-Omniscience: non-hallucinationnot in index	19.5%#287	Artificial Analysis

Instruction following

Benchmark	thinking	Source	Trend
IFBench	54.6%#151	Artificial Analysis

Long context

Benchmark	thinking	Source	Trend
AA-LCR	21.7%#425	Artificial Analysis

Composite

Benchmark	thinking	Source	Trend
AA Intelligence Index v4.3.2	9.6#393	Artificial Analysis

New this month

Questions people ask

What are Tri-21B-Think's benchmark scores?: As of 25 Oct 2026, Tri-21B-Think (thinking effort) scores 43.4 ±5.9 on the BenchLeader Index, strongest in instruction following (55) and weakest in knowledge (33). That rests on 6 independent benchmark results, and every score on this page links to its publisher.
How does Tri-21B-Think rank?: Tri-21B-Think (thinking effort) is #568 of 785 ranked configurations on 25 Oct 2026. 350 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
When was Tri-21B-Think released?: Trillion Labs released Tri-21B-Think on 10 Feb 2026.

Data as of 25 Oct 2026. Compare with another model.

Cite as: BenchLeader, “Tri-21B-Think: benchmarks, pricing, speed and rank”, data as of 25 Oct 2026, https://www.benchleader.com/models/tri-21b.